A method and device for ultra-short-term prediction of equipment parameters of a comprehensive energy system

Abstract

The present invention discloses a method and device for ultra-short-term prediction of equipment parameters in a comprehensive energy system. By building a historical database, mechanism analysis and correlation analysis are completed to provide a theoretical basis for the selection of characteristic parameters in the process of establishing a prediction model. By dividing the historical database according to different equipment operating conditions and calculating their corresponding characteristic identification quantities, a discrimination criterion is provided for subsequent operating condition prediction, searching and obtaining the input data of the prediction model. Finally, through similarity analysis and grey degree modeling, ultra-short-term prediction of prediction parameters is realized, and prediction results of prediction parameters in the next 0-4 hours with a time resolution of 5 minutes are provided. The present invention effectively improves the prediction accuracy and provides an effective reference basis for the optimal scheduling and economic operation of the regional comprehensive energy system.

Description

Technical Field

[0001] The present invention relates to a method and device for ultra-short-term prediction of equipment parameters in an integrated energy system, and belongs to the technical field of integrated energy prediction. Background Technique

[0002] The integrated energy system is an important technical means to improve energy utilization efficiency and reduce energy consumption costs. Accurate supply and demand prediction is the key foundation and important prerequisite for the economic planning, collaborative management, and optimal scheduling of the integrated energy system. To achieve accurate supply and demand prediction, the ultra-short-term prediction of equipment parameters in the integrated energy system is crucial. The ultra-short-term prediction of equipment parameters needs to follow the inherent change laws of the parameters themselves and take into account the various impacts of external factors on parameter changes in order to accurately predict the changes in equipment parameters within the next 0-4 hours.

[0003] At present, significant progress has been made in the basic theory of the ultra-short-term prediction research on equipment parameters in the integrated energy system, and certain achievements have also been obtained in engineering practical applications. Generally speaking, it can be divided into two categories. One is the prediction method based on the mechanism model, which relies on the mathematical description of the equipment, has clear physical meaning and clear functional relationships, and has strong extrapolation performance in the prediction process. However, as the complexity of the equipment mechanism increases, the difficulty of its mathematical description increases sharply, and the feasibility of prediction implementation and the reliability of prediction results also decrease sharply. The other is the prediction method based on the data model, which is characterized by extracting the correlation relationship between variables based on data samples, having a short development cycle, small computing power requirements, and relatively low comprehensive cost for prediction solution. However, similarly, this method usually has difficulty in explaining its decision-making basis and operation logic, and the uncertainty of the reliability of prediction results and potential safety risks limit its popularization and application in the industrial field.

[0004] Overall, due to the large variety of equipment involved in the integrated energy system, and the fluctuations in the operation laws of different equipment in different time and space spans, combined with the uncertainty of various external factor influences, the prediction algorithms based on a single model of the mechanism model or the data model have poor effects.

[0005] Based on this, in order to improve the overall performance of the ultra-short-term prediction method for integrated energy system equipment and enhance the application effect of the method, the present invention proposes a method and device for ultra-short-term prediction of equipment parameters in the integrated energy system. Summary of the Invention

[0006] Objective: To overcome the deficiencies in the existing technologies, the present invention provides a method and device for ultra-short-term prediction of equipment parameters of an integrated energy system. By building a historical database, mechanism analysis and correlation analysis are completed to provide a theoretical basis for the selection of characteristic parameters in the process of establishing a prediction model. By dividing the historical database according to different equipment operating conditions and calculating the corresponding characteristic identification quantities, a discrimination criterion is provided for subsequent operating condition prediction, searching, and obtaining the input data of the prediction model. Finally, through similarity analysis and grey-scale modeling, ultra-short-term prediction of prediction parameters is realized, and prediction results of prediction parameters in the future 0-4 hours with a time resolution of 5 minutes are provided.

[0007] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0008] The present invention provides a method for ultra-short-term prediction of equipment parameters of an integrated energy system, including the following steps:

[0009] Collect historical time-series data of each equipment parameter including the prediction parameter in the integrated energy system equipment, and construct a historical database.

[0010] Select characterization parameters related to the prediction parameter from the equipment parameters in the historical database according to the mechanism analysis of the equipment.

[0011] Conduct correlation analysis on the characterization parameters, and extract characteristic parameters from the characterization parameters.

[0012] Calculate characteristic identification quantities under different operating condition categories according to the characteristic parameters.

[0013] Construct an on-line prediction model for the integrated energy system equipment.

[0014] Determine the operating condition category of the equipment to be predicted, select the extended sequence of the original basic sequence of the characteristic parameter most similar to the current moment from the corresponding historical database as the input of the on-line prediction model for the integrated energy system equipment, solve the on-line prediction model for the integrated energy system equipment, and obtain the predicted value of the equipment parameter of the equipment to be predicted.

[0015] An ultra-short-term prediction device for equipment parameters of an integrated energy system includes the following modules:

[0016] A data acquisition module, configured to collect historical time-series data of each equipment parameter including the prediction parameter in the integrated energy system equipment, and construct a historical database.

[0017] A mechanism analysis module, configured to select characterization parameters related to the prediction parameter from the equipment parameters in the historical database according to the mechanism analysis of the equipment.

[0018] A correlation analysis module, configured to conduct correlation analysis on the characterization parameters, and extract characteristic parameters from the characterization parameters.

[0019] The operating condition analysis module is used to calculate the characteristic identification quantity under different operating condition categories according to the characteristic parameters.

[0020] The online prediction model construction module is used to construct an online prediction model for the integrated energy system equipment.

[0021] The real-time prediction module is used to determine the operating condition category of the equipment to be predicted, select the extended sequence of the original basic sequence of the characteristic parameters that is most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, solve the online prediction model for the integrated energy system equipment, and obtain the predicted value of the equipment parameters of the equipment to be predicted.

[0022] As a preferred solution, it further includes: when there are bad points in the historical database, taking the bad points as the center, selecting the normal data before and after the bad points in chronological order for curve fitting, interpolating and calculating the values at the bad points according to the fitting function, and using them as the correction values of the bad points.

[0023] As a preferred solution, the correlation analysis of the characterization parameters and the extraction of characteristic parameters from the characterization parameters include the following steps:

[0024] 1) Obtain the reference sequence of the prediction parameter x 0 and the comparison sequences of p characterization parameters from the historical database.

[0025] x 0 ={x 0 (1), x 0 (2), …, x 0 (i), …, x 0 (n)}

[0026]

[0027] In the formula, x 0 is the reference sequence, x 0 (i) is the i-th related parameter of the prediction parameter, is the comparison sequence of the i-th characterization parameter, is the j-th related parameter of the i-th comparison sequence, n is the number of elements contained in the sequence, and p is the number of characterization parameters.

[0028] 2) Perform dimensionless initialization processing on the reference sequence and the comparison sequences to obtain the dimensionless initialized reference sequence and comparison sequences.

[0029] x' 0 ={1, x 0 (2) / x 0 (1), …, x 0 (n) / x 0(1)}

[0030]

[0031] 3) Calculate the reference sequence x' after dimensionless initialization 0 and the comparison sequence of the grey correlation coefficient on the k-th correlation parameter:

[0032]

[0033] where η is the resolution coefficient, represents the two-level minimum difference, represents the two-level maximum difference.

[0034] 4) Calculate the grey correlation degree γ of each characterization parameter i .

[0035]

[0036] 5) Set the grey correlation degree threshold γ 0 , and extract from it the characterization parameters {γ i |γ i >γ 0 , i = 1, 2, …, p} to form a new characteristic parameter sequence x 1 , x 2 , …, x q , where q is the number of characteristic parameters, q ≤ p.

[0037] As an optimal solution, calculating the characteristic identification quantity under different operating condition categories according to the characteristic parameters includes the following steps:

[0038] Divide the operating condition categories according to the operating conditions of the equipment, and divide the historical data according to the operating condition categories.

[0039] According to the divided historical data, calculate the characteristic identification quantity corresponding to the operating condition category. It includes the following steps:

[0040] 1) Establish a regression model 1 of the characteristic parameters x 2 , …, x q for the prediction parameter x 0 :

[0041]

[0042] where is the regression coefficient, is the object estimated value calculated based on the regression model.

[0043] 2) Obtain each regression coefficient from the following formula

[0044]

[0045] Wherein, are the average values of x 0 , x 1 , x 2 , …, x q respectively, and T and R are the deviation matrix and the regression matrix of the regression model; there is

[0046]

[0047]

[0048] Wherein, T ij represents the element in the i-th row and j-th column of matrix T, and R i represents the i-th element in matrix R.

[0049] 3) Use λ to characterize the overall correlation degree of the prediction parameter x 0 and the characteristic parameters x 1 , x 2 , …, x q under this working condition, and its calculation formula is:

[0050]

[0051] Take λ as the characteristic identification quantity corresponding to each group of working condition data.

[0052] As an optimized scheme, the construction of the on-line prediction model for the integrated energy system equipment includes the following steps:

[0053] 1) When the prediction period arrives, taking the current moment as the starting point, simultaneously trace back m historical data of the parameters x 0 , x 1 , x 2 ,...,, x q as the basic original sequence for on-line modeling:

[0054]

[0055] Perform a first-order accumulation generation (1-AGO) on to obtain the sequence

[0056]

[0057] Wherein,

[0058] 2) Generate the adjacent mean sequence from

[0059] Wherein,​

[0060] In the formula, α is the background value.

[0061] 3) Establish the gray degree prediction model as

[0062]

[0063] In the formula, a is the development coefficient, b i is the driving coefficient, h 1 (k - 1) is the linear correction amount; h 2 is the gray action amount.

[0064] 4) The gray degree prediction model can be transformed into

[0065]

[0066] In the formula,

[0067] 5) Perform a first-order accumulated subtraction reduction operation on the above formula to obtain the online prediction model of the prediction parameters:

[0068]

[0069] As an optimized solution, the specific calculation expressions of the parameters a, b i , h 1 (k - 1), h 2 are:

[0070] U = (B T B) -1 B T Y

[0071] In the formula, B T represents the transpose matrix of matrix B, (B T B) -1 represents the inverse matrix of matrix B T B

[0072]

[0073] The determination of the operating condition category of the device to be predicted includes the following steps:

[0074] Using the original basic sequence, calculate the prediction parameter x 0 and the characteristic parameters x 1 , x 2 ,..., x q and the real-time identification quantity λ at the current moment.

[0075] Determine the operating condition category of the equipment to be predicted according to the real-time identification quantity λ of the current operating condition of the equipment and the determination condition. The determination condition is to satisfy the value of l below, where L is the total number of operating condition classifications.

[0076] Select the extended sequence of the original basic sequence of the characteristic parameters that is most similar to the current moment from the corresponding historical database as the input of the online prediction model of the integrated energy system equipment, solve the online prediction model of the integrated energy system equipment, and obtain the predicted value of the equipment parameters of the equipment to be predicted, including the following steps:

[0077] 1) In the corresponding historical operating condition data sequence, starting from the first historical data of the prediction parameter, select m data points to form a historical sequence Through the measured sequence of the prediction parameter Perform similarity judgment to find the starting serial number point s of the similar sequence in the historical data sequence that satisfies the following formula;

[0078]

[0079] In the formula, is the data of the (s + k - 1)-th point in the historical sequence of the prediction parameter from the s-th point to the (s + m - 1)-th point;

[0080] 2) Obtain the corresponding historical sequence of characteristic parameters

[0081]

[0082] In the formula, is the data of the (s + m + k - 1)-th point in the historical sequence of characteristic parameters from the (s + m)-th point to the (s + m + f - 1)-th point, k is the number of points to be predicted, and f is the total number of prediction points;

[0083] Merge into the original basic sequence of characteristic parameters to obtain the extended sequence of the original basic sequence of characteristic parameters

[0084]

[0085] Perform a first-order accumulation generation (1-AGO) on to obtain the input sequence of the prediction model

[0086]

[0087] In the formula,

[0088] 3) Merge Replace the following formula Calculate

[0089]

[0090] 4) Substitute into the following formula, and the predicted values of the equipment prediction parameters at the next f points are obtained as follows:

[0091]

[0092] Beneficial effects: Compared with the prior art, the short-term prediction method and device for equipment parameters of an integrated energy system provided by the present invention have the following beneficial effects: By integrating the theoretical accuracy of the mechanism model and the strong generalization and in-depthness of the data-driven model, and by means of correlation analysis, operating condition data division, and similarity principle, an online gray-scale short-term prediction model is finally established to realize the short-term prediction of specific parameters of equipment in the integrated energy system under the condition of multi-factor coupling, effectively improving the prediction accuracy and providing an effective reference basis for the optimal scheduling and economic operation of the regional integrated energy system. Brief description of the drawings

[0093] Figure 1 is the flowchart of the implementation steps of the method of the present invention.

[0094] Figure 2 is the structural schematic diagram of the device of the present invention. Detailed implementation manners

[0095] The present invention will be further described in detail below with reference to specific embodiments.

[0096] A short-term prediction method and device for equipment parameters of an integrated energy system, as Figure 1 shown. On the one hand, the present invention provides a short-term prediction method for equipment parameters of an integrated energy system, including the following steps:

[0097] Collect the historical time-series data of each equipment parameter including the prediction parameter in the equipment of the integrated energy system, and construct a historical database.

[0098] Select the characteristic parameters related to the prediction parameter from the equipment parameters in the historical database according to the mechanism analysis of the equipment.

[0099] Conduct correlation analysis on the characteristic parameters, and extract the feature parameters from the characteristic parameters.

[0100] Calculate the characteristic identification quantity under different operating condition categories according to the feature parameters.

[0101] Construct an online prediction model for the equipment of the integrated energy system.

[0102] Determine the operating condition category of the device to be predicted, select the extended sequence of the original basic sequence of characteristic parameters that is most similar to the current moment from the corresponding historical database as the input of the on-line prediction model of the integrated energy system device, solve the on-line prediction model of the integrated energy system device, and obtain the predicted value of the device parameters of the device to be predicted.

[0103] When there are bad points in the historical database, including data missing and gross error, the method of curve fitting is used for data correction. Specifically, with the bad point as the center, the normal data before and after the bad point are selected in chronological order for curve fitting, and the value at the bad point is calculated by interpolation according to the fitting function and used as the correction value of the bad point.

[0104] The mechanism analysis of the device lies in using a mechanism model with clear physical or practical significance to find the variation law of the prediction parameter with the characterization parameter. Taking the generator set as an example, the mechanism models involved in the device operation process include:

[0105] 1) Energy balance equation

[0106]

[0107] 2) Mass balance equation

[0108]

[0109] 3) Working medium characteristic equation

[0110] f(h, p, T) = 0

[0111] 4) Boundary constraint equation

[0112] f(λ 1 , λ 2 , λ 3 , …) = 0

[0113] In the formula: ρ represents density, c represents specific heat capacity, V represents volume, T represents temperature, τ represents time, q in represents input energy, q out represents output energy, D in represents inlet flow rate, D out represents outlet flow rate, h represents enthalpy value, P represents pressure, λ 1 , λ 2 , λ 3 represents constraint conditions.

[0114] Therefore, when the output energy of the unit is used as the prediction parameter, the relevant characterization parameters can be selected as temperature, pressure and flow rate. After the characterization parameters are determined, the characteristic parameters are extracted from them through correlation analysis.

[0115] Performing correlation analysis on the characterization parameters and extracting characteristic parameters from the characterization parameters includes the following steps:

[0116] 1) Obtain the reference sequence of the prediction parameter x 0 and the comparison sequences of p characterization parameters from the historical database.

[0117] x 0 ={x 0 (1), x 0 (2), … x 0 (i) …, x 0 (n)}

[0118]

[0119] In the formula, x 0 is the reference sequence, x 0 (i) is the i-th relevant parameter of the prediction parameter, is the comparison sequence of the i-th characterization parameter, is the j-th relevant parameter of the i-th comparison sequence, n is the number of elements included in the sequence, and p is the number of characterization parameters.

[0120] 2) Further, perform dimensionless initialization processing on the reference sequence and the comparison sequences to obtain the dimensionless initialized reference sequence and comparison sequences.

[0121] x′ 0 ={1, x 0 (2) / x 0 (1), …, x 0 (n) / x 0 (1)}

[0122]

[0123] 3) Calculate the grey correlation coefficient between the dimensionless initialized reference sequence x′ 0 and the comparison sequence on the k-th relevant parameter:

[0124]

[0125] In the formula, η is the resolution coefficient, η ∈ [0, 1], and when η ≤ 0.55, the resolution ability is the best.

[0126] represents the two-level minimum difference, represents the two-level maximum difference.

[0127] 4) Further, calculate the grey correlation degree γ i of each characterization parameter.

[0128]

[0129] 5) Set the grey correlation degree threshold γ 0 = 0.6, and extract the characterization parameters {γ i |γ i > γ 0 , i = 1, 2,..., p} to form a new sequence of characteristic parameters x 1 , x 2 ,..., x q , where q is the number of characteristic parameters, and q ≤ p.

[0130] Calculating the characteristic identification quantity under different operating condition categories according to the characteristic parameters includes the following steps:

[0131] Divide the operating conditions of the equipment into eight operating condition categories: full state, 90% state, 70% state, 50% state, 30% state, 10% state, rising state, and falling state, and divide the historical data according to the operating condition categories.

[0132] According to the divided historical data, calculate the characteristic identification quantity corresponding to the operating condition category. It includes the following steps:

[0133] 1) Establish a regression model of the characteristic parameters x 1 , x 2 ,..., x q for the prediction parameter x 0

[0134]

[0135] In the formula, is the regression coefficient, is the object estimated value calculated based on the regression model.

[0136] 2) Further, each regression coefficient can be obtained by the following formula

[0137]

[0138] In the formula, are the averages of x 0 , x 1 , x 2 ,..., x q respectively, and T and R are the deviation matrix and regression matrix of the regression model, and there are

[0139]

[0140]

[0141] In the formula, T ij ​Denote the element in the \(i\)-th row and \(j\)-th column of matrix \(T\), \(R\) i Denote the \(i\)-th element in matrix \(R\).

[0142] 3) Further, use \(\lambda\) to characterize the prediction parameter \(x\) 0 and the feature parameter \(x\) 1 , \(x\) 2 , \(\cdots\), \(x\) q The overall correlation degree under this working condition, and its calculation formula is:

[0143]

[0144] Thus, take \(\lambda\) as the feature identification quantity corresponding to each group of working condition data. When obtaining the measured data of \(x\) 0 , \(x\) 1 , \(x\) 2 , \(\cdots\), \(x\) q , the current working condition state of the equipment can be predicted by calculating its real-time identification quantity and comparing it with the feature identification quantity, further providing a basis for the selection of prediction samples for the implementation of real-time prediction.

[0145] The construction of the online prediction model for integrated energy system equipment includes the following steps:

[0146] When the prediction period arrives, starting from the current moment, simultaneously trace back \(m\) historical data of the parameter \(x\) 0 , \(x\) 1 , \(x\) 2 , \(\cdots\), \(x\) q as the basic original sequence for online modeling, and establish a prediction model:

[0147] 1) Basic original sequence

[0148]

[0149] Perform a first-order accumulation generation (1-AGO) on to obtain the sequence

[0150]

[0151] wherein,

[0152] 2) Generate the adjacent mean sequence from

[0153] wherein,

[0154] where \(\alpha\) is the background value, and by default, it is taken as 0.5.

[0155] 3) Further, establish the grey prediction model as ​

[0156]

[0157] In the formula, a is the development coefficient, and b i is the driving coefficient, and h 1 (k - 1) is the linear correction amount; h 2 is the grey action amount. The parameters a, b i , h 1 (k - 1), h 2 The specific calculation expressions are as follows:

[0158] U = (B T B) -1 B T Y

[0159] In the formula, B T represents the transpose matrix of matrix B, and (B T B) -1 represents the inverse matrix of matrix B T B

[0160]

[0161] 4) Further, the grey degree prediction model can be transformed into

[0162]

[0163] In the formula,

[0164] 5) Perform a first-order accumulated subtraction reduction operation on the above formula to obtain the online prediction model of the prediction parameters:

[0165]

[0166] Determining the operating condition category of the device to be predicted includes the following steps:

[0167] Using the original basic sequence, calculate the prediction parameter x 0 and the characteristic parameter x 1 , x 2 ,..., x q The real-time identification quantity λ at the current moment.

[0168] According to the real-time identification quantity λ of the current operating condition of the device and the determination condition, determine the operating condition category of the device to be predicted. The determination condition is to satisfy the value of l for l = 1, 2,..., L, where L is the total number of operating condition classifications.

[0169] Select the extended sequence of the feature parameter original basic sequence that is most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, and solve the online prediction model for the integrated energy system equipment to obtain the predicted value of the equipment parameters of the equipment to be predicted, including the following steps:

[0170] 1) In the corresponding historical operating condition data sequence, starting from the first historical data of the prediction parameter, select m data points to form a historical sequence Through the measured sequence of the prediction parameter Perform similarity judgment to find the starting serial number point s of the similar sequence in the historical data sequence that satisfies the following formula;

[0171]

[0172] In the formula, is the data of the (s + k - 1)th point in the historical sequence of the prediction parameter from the sth point to the (s + m - 1)th point;

[0173] 2) Obtain the corresponding historical sequence of feature parameters

[0174]

[0175] In the formula, is the data of the (s + m + k - 1)th point in the historical sequence of feature parameters from the (s + m)th point to the (s + m + f - 1)th point, k is the number of points to be predicted, and f is the total number of predictions;

[0176] Merge into the original basic sequence of feature parameters to obtain the extended sequence of the original basic sequence of feature parameters

[0177]

[0178] Perform a first-order accumulation generation (1-AGO) on to obtain the input sequence of the prediction model

[0179]

[0180] In the formula,

[0181] 3) Replace in the following formula Calculate

[0182]

[0183] 4) Replace Substituting into the following formula, the predicted values of the device prediction parameters at the next f points are obtained as follows:

[0184]

[0185] As Figure 2 shown, in a second aspect, the present invention provides a very short-term prediction device for the parameters of an integrated energy system device, including the following modules:

[0186] A data acquisition module, configured to acquire historical time-series data of each device parameter including prediction parameters in the integrated energy system device, and construct a historical database.

[0187] A mechanism analysis module, configured to select characterization parameters related to the prediction parameters from the device parameters in the historical database according to the mechanism analysis of the device.

[0188] An association analysis module, configured to perform association analysis on the characterization parameters and extract feature parameters from the characterization parameters.

[0189] An operating condition analysis module, configured to calculate characteristic identification quantities under different operating condition categories according to the feature parameters.

[0190] An online prediction model construction module, configured to construct an online prediction model for the integrated energy system device.

[0191] A real-time prediction module, configured to determine the operating condition category of the device to be predicted, select an extended sequence of the original basic sequence of the feature parameters most similar to the current moment from the corresponding historical database as the input of the online prediction model of the integrated energy system device, solve the online prediction model of the integrated energy system device, and obtain the predicted value of the device parameter of the device to be predicted.

[0192] Example 1:

[0193] Taking a blast furnace in an energy device in an integrated energy system as an example, the parameters of the gas generation amount of the blast furnace are predicted. The specific steps are as follows:

[0194] Step 1, acquire historical time-series data of the parameters of the blast furnace in the integrated energy system and construct a historical database.

[0195] Obtain historical time-series data of all device parameters including the gas generation amount in the blast furnace device of the integrated energy system, and perform data correction on possible bad points in the historical time-series data, including data missing or data gross error, by using the method of curve fitting. The specific process includes: taking the bad point as the center, selecting normal data before and after the bad point in chronological order for curve fitting, interpolating and calculating the value at the bad point according to the fitting function, and using it as the correction value of the bad point.

[0196] Step 2: Based on the mechanism analysis of the blast furnace, determine the characterization parameters related to the blast furnace gas generation volume.

[0197] Select reasonable characterization parameters according to the analysis results of the blast furnace gas generation mechanism. During the blast furnace smelting process, materials such as iron ore, coke, and limestone are charged from the top of the furnace, and after undergoing reduction reactions with preheated air (mixed with oxygen enrichment and pulverized coal) inside the blast furnace, the blast furnace gas is finally discharged from the top of the furnace. Therefore, when predicting the parameters of the gas generation volume, the relevant characterization parameters can be selected as coke ratio, coal ratio, oxygen enrichment, air volume, air temperature, air pressure, and furnace differential pressure.

[0198] Step 3: Conduct correlation analysis on the characterization parameters to extract the characteristic parameters related to the blast furnace gas generation volume.

[0199] After the characterization parameters are determined, extract the characteristic parameters from them through correlation analysis as the basis for the subsequent calculation of the operating condition characteristic identifier and the establishment of the prediction model. The specific implementation steps are as follows:

[0200] 1) Obtain the reference sequence of the gas generation volume and the comparison sequences of coke ratio, coal ratio, oxygen enrichment, air volume, air temperature, air pressure, and furnace differential pressure from the historical database.

[0201] x 0 ={x 0 (1), x 0 (2), … x 0 (i) …, x 0 (n)}

[0202]

[0203] In the formula, x 0 is the reference sequence, x 0 (i) is the i-th relevant parameter of the gas generation volume, is the comparison sequence of the i-th characterization parameter, is the j-th relevant parameter of the i-th comparison sequence, n is the number of elements contained in the sequence, and p is the number of characterization parameters.

[0204] 2) Further, perform dimensionless initialization processing on the reference sequence and the comparison sequences to obtain the dimensionless initialized reference sequence and comparison sequences.

[0205] x′ 0 ={1, x 0 (2) / x 0 (1), …, x 0 (n) / x 0 (1)}

[0206]

[0207] 3) Calculate the dimensionless initialized reference sequence x′ 0 and the comparison sequence on the k-th correlation parameter: the grey correlation coefficient

[0208]

[0209] where η is the resolution coefficient, η ∈ [0, 1], and the resolution ability is optimal when η ≤ 0.55.

[0210] represents the two-level minimum difference, represents the two-level maximum difference.

[0211] 4) Further, calculate the grey correlation degree γ of each characterization parameter i .

[0212]

[0213] 5) Set the grey correlation degree threshold γ 0 = 0.6, and extract the characterization parameters {γ i |γ i > γ 0 , i = 1, 2,..., p} to form a new characteristic parameter sequence x 1 , x 2 ,..., x q , where q is the number of characteristic parameters, q ≤ p.

[0214] Among the calculated grey correlation degrees, air volume > coke ratio > blast pressure > coal ratio > blast temperature > 0.6 > oxygen enrichment > furnace differential pressure. Therefore, air volume, coke ratio, blast pressure, coal ratio, and blast temperature are used as the characteristic parameters for predicting the blast furnace gas generation volume.

[0215] Step 4, calculate the characteristic identification quantity of the blast furnace gas generation volume under different operating condition categories.

[0216] After determining the characteristic parameters, according to the operating conditions of the blast furnace equipment, the operating conditions are divided into eight operating condition categories: full load, 90% load, 70% load, 50% load, 30% load, 10% load, load increase, and load decrease. And the historical data of the blast furnace gas generation volume and the characteristic parameters are divided by the operating condition categories, and the corresponding characteristic identification quantities are calculated. The implementation steps include:

[0217] 1) Establish a regression model of air volume, coke ratio, blast pressure, coal ratio, and blast temperature on the gas generation volume x 0

[0218]

[0219] where is the regression coefficient, ​is the estimated value of the prediction parameter calculated based on the regression model, x 1 , x 2 ,..., x q are the characteristic parameters of the air volume, coke ratio, blast pressure, coal ratio, and blast temperature with q = 5.

[0220] 2) Further, each regression coefficient can be obtained from the following formula

[0221]

[0222] In the formula, are the average values of x 0 , x 1 , x 2 ,..., x q respectively. T and R are the deviation matrix and regression matrix of the regression model, and there are

[0223]

[0224]

[0225] In the formula, T ij represents the element in the i-th row and j-th column of matrix T, and R i represents the i-th element in matrix R.

[0226] 3) Further, use λ to characterize the overall correlation degree between the gas generation amount and the air volume, coke ratio, blast pressure, coal ratio, and blast temperature under this working condition. Its calculation formula is:

[0227]

[0228] Thus, repeat this step to calculate the λ value of each group of working condition data and use it as the characteristic identification quantity corresponding to each working condition. When the measured data of the gas generation amount, air volume, coke ratio, blast pressure, coal ratio, and blast temperature are obtained, the current working condition state of the blast furnace can be judged by calculating its real-time characteristic identification quantity and comparing it with the characteristic identification quantity, which further provides a basis for the selection of prediction samples for the implementation of real-time prediction.

[0229] Step 5, Establishment of the online prediction model

[0230] When the prediction period arrives, starting from the current moment, simultaneously trace back m historical data of the gas generation amount, air volume, coke ratio, blast pressure, coal ratio, and blast temperature as the basic original sequence for online modeling. The process of establishing the prediction model includes:

[0231] 1) Obtain the basic original sequences of the gas generation amount, coke ratio, coal ratio, air volume, blast pressure, and blast temperature parameters

[0232]

[0233] Do One cumulative generation (1-AGO) to obtain the sequence

[0234]

[0235] In the formula,

[0236] 2) For the blast furnace gas generation volume, from Generate the adjacent mean sequence

[0237] In the formula,

[0238] In the formula, α is the background value, and the default value is 0.5.

[0239] 3) Further, establish the grey prediction model as

[0240]

[0241] In the formula, a is the development coefficient, b i Is the driving coefficient, h 1 (k - 1) is the linear correction amount; h 2 Is the grey action amount. The parameters a, b i , h 1 (k - 1), h 2 The specific calculation expressions are as follows:

[0242] U=(B T B) -1 B T Y

[0243] In the formula, B T Represents the transpose matrix of matrix B, (B T B) -1 Represents the inverse matrix of matrix B T

[0244]

[0245] 4) Further, the grey prediction model can be transformed into

[0246]

[0247] In the formula,

[0248] 5) Do one subtraction reduction operation on the above formula to obtain the grey prediction value of the blast furnace gas generation volume as:

[0249]

[0250] ​Step 6, Prediction Input and Prediction Output

[0251] Using the original basic sequence in Step 5, Based on Step 4, calculate the real-time identification quantity λ of the gas generation amount, air volume, coke ratio, blast pressure, coal ratio, and blast temperature at the current moment. Further, the determination condition for the current working condition of the equipment is satisfied for the value of l below, where L is the total number of working condition classifications.

[0252] After determining the current working condition state of the equipment, it is necessary to find the extended sequence of the original basic sequence of the characteristic parameters that is most similar to the current moment from the corresponding historical working condition data sequence as the input of the prediction model. The specific implementation process is as follows:

[0253] 1) In the corresponding historical working condition data sequence, starting from the first historical data of the prediction parameter, select m data points to form a historical sequence Through comparison with the measured sequence of the prediction parameter perform a similarity judgment to find the starting serial number point s of the similar sequence that satisfies the following formula in the historical data sequence;

[0254]

[0255] In the formula, is the data of the (s + k - 1)-th point in the historical sequence of the prediction parameter from the s-th point to the (s + m - 1)-th point;

[0256] 2) Further, obtain the corresponding historical sequence of characteristic parameters

[0257]

[0258] In the formula, is the data of the (s + m + k - 1)-th point in the historical sequence of characteristic parameters from the (s + m)-th point to the (s + m + f - 1)-th point, k is the number of points to be predicted, and f is the total number of prediction points;

[0259] Merge into the original basic sequence of characteristic parameters to obtain the extended sequence of the original basic sequence of characteristic parameters

[0260]

[0261] Perform a first-order accumulation generation (1-AGO) on to obtain the input sequence of the prediction model

[0262]

[0263] In the formula,

[0264] 3) Replace in the following formula and calculate

[0265]

[0266] 4) Substitute into the following formula, and finally obtain the predicted values of the predicted parameters at a total of f points with a time resolution of 5 minutes in the future 0 - 4 hours as:

[0267]

[0268] Embodiment 2:

[0269] The present invention provides an implementation method and system for the ultra - short - term prediction of integrated energy system equipment. Through the construction of a historical database, characteristic parameters are obtained based on mechanism analysis for correlation analysis and feature parameter extraction. The operating condition data is divided in combination with the operating conditions of energy equipment to build a sample space. Subsequently, an online grey - degree ultra - short - term prediction model is established using the measured data of the predicted parameters and feature parameters, and the input of the online prediction model is obtained through the prediction of the real - time operating conditions. Finally, the ultra - short - term prediction of the predicted parameters is realized. The model takes into account the theoretical accuracy of the mechanism model and the strong generalization and in - depth nature of the data - driven model, can realize the ultra - short - term prediction of integrated energy equipment under multi - factor coupling conditions, effectively improve the prediction accuracy, and provide an effective reference basis for the optimal dispatching and economic operation of the regional integrated energy system.

[0270] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for ultra-short-term prediction of equipment parameters in an integrated energy system, characterized in that: It includes the following steps: Collect historical time-series data of each equipment parameter including the prediction parameter in the integrated energy system equipment, and construct a historical database; Select characterization parameters related to the prediction parameter from the equipment parameters in the historical database according to the mechanism analysis of the equipment; Conduct correlation analysis on the characterization parameters, and extract characteristic parameters from the characterization parameters; Calculate characteristic identification quantities under different operating condition categories according to the characteristic parameters; Construct an online prediction model for the integrated energy system equipment; Determine the operating condition category of the equipment to be predicted, select the extended sequence of the original basic sequence of the characteristic parameter most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, solve the online prediction model for the integrated energy system equipment, and obtain the predicted value of the equipment parameter of the equipment to be predicted; The correlation analysis of the characterization parameters to extract characteristic parameters from the characterization parameters includes the following steps: 1) Obtain the prediction parameter x from the historical database 0 for the reference sequence and the comparison sequences of p characterization parameters x 0 = {x 0 (1), x 0 (2), … x 0 (i) …, x 0 (n)} where x 0 is the reference sequence, x 0 (i) is the i-th relevant parameter of the prediction parameter, is the comparison sequence of the i-th characterization parameter, is the j-th relevant parameter of the i-th comparison sequence, n is the number of elements included in the sequence, and p is the number of characterization parameters; 2) Perform dimensionless initialization processing on the reference sequence and the comparison sequence to obtain the dimensionless initialized reference sequence and comparison sequence; x' 0 ={1, x 0 (2) / x 0 (1), …, x 0 (n) / x 0 (1)} 3) Calculate the reference sequence x' after dimensionless initialization 0 and the comparison sequence on the k-th correlation parameter: where η is the resolution coefficient, represents the minimum difference between two levels, represents the maximum difference between two levels; 4) Calculate the grey relational grade γ of each characteristic parameter i ; 5) Set the grey relational degree threshold γ 0 , and extract {γ i |γ i >γ 0 , i = 1, 2, …, p} of the characterization parameters to form a new sequence of characteristic parameters x 1 , x 2 , …, x q , where q is the number of characteristic parameters, q ≤ p; The calculation of the characteristic identification quantity under different operating condition categories according to the characteristic parameters includes the following steps: Divide the operating condition categories according to the operating conditions of the equipment, and divide the historical data according to the operating condition categories; According to the divided historical data, calculate the characteristic identification quantity corresponding to the operating condition category; including the following steps: 1) Establish the characteristic parameter x 1 , x 2 , …, x q For the regression model of the prediction parameter x 0 ​ In the formula, is the regression coefficient, is the estimated value of the object calculated based on the regression model; 2) Obtain each regression coefficient from the following formula In the formula, are respectively the average values of x 0 , x 1 , x 2 , …, x q ; T and R are respectively the deviation matrix and the regression matrix of the regression model; where T ij represents the element in the \(i\)-th row and \(j\)-th column of matrix \(T\), and R i represents the \(i\)-th element in matrix \(R\); 3) Use λ to characterize the prediction parameter x 0 With the characteristic parameter x 1 , x 2 , …, x q The overall correlation degree under this working condition, and its calculation formula is: Take λ as the characteristic identification quantity corresponding to each group of operating condition data; The construction of the online prediction model for the integrated energy system equipment includes the following steps: 1) When the prediction period is reached, starting from the current moment, simultaneously for the parameters x 0 , x 1 , x 2 , …, x q Trace back m historical data as the basic original sequence for online modeling: For perform an accumulation generation to obtain a sequence In the formula, 2) Generated by Generate the adjacent mean sequence In the formula, In the formula, α is the background value; 3) Establish a gray prediction model as where a is the development coefficient, b i is the driving coefficient, h 1 (k - 1) is the linear correction amount; h 2 is the grey action amount; 4) The gray prediction model can be transformed into In the formula, 5) Perform a first-order accumulated reduction operation on the above formula to obtain the online prediction model of the prediction parameter: The determination of the operating condition category of the equipment to be predicted includes the following steps: Using the original basic sequence, calculate the prediction parameter x 0 and the characteristic parameter x 1 , x 2 , …, x q and the real-time identification quantity λ at the current moment; Determine the operating condition category of the equipment to be predicted according to the real-time identification quantity λ of the current working condition of the equipment and the determination condition; the determination condition is to satisfy the value of l below, where L is the total number of working condition classifications; The selection of the extended sequence of the original basic sequence of the characteristic parameter most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, solving the online prediction model for the integrated energy system equipment, and obtaining the predicted value of the equipment parameter of the equipment to be predicted includes the following steps: 1) In the corresponding historical operating condition data sequence, starting from the first historical data of the prediction parameter, select m data points to form a historical sequence By comparing with the measured sequence of the prediction parameter Perform similarity judgment to find the starting serial number point s of the similar sequence that satisfies the following formula in the historical data sequence; wherein, is the data of the (s + k - 1)-th point in the historical sequence of prediction parameters starting from the s-th point and ending at the (s + m - 1)-th point; 2) Obtain the corresponding historical sequence of characteristic parameters Wherein, is the data of the (s + m + k - 1)-th point in the historical sequence of characteristic parameters starting from the (s + m)-th point to the (s + m + f - 1)-th point, k is the number of points to be predicted, and f is the total number of predicted points; Incorporate into the original basic sequence of characteristic parameters to obtain an extended sequence of the original basic sequence of characteristic parameters For perform an accumulation generation to obtain the input sequence of the prediction model In the formula, 3) Substitute in the following formula and calculate 4) Substitute into the following formula, and the predicted values of the device prediction parameters at the next f points are obtained as:

2. A method for ultra-short-term prediction of equipment parameters in an integrated energy system according to claim 1, characterized in that: It further includes: When there are bad points in the historical database, take the bad point as the center, select the normal data before and after the bad point in chronological order for curve fitting, interpolate and calculate the value at the bad point according to the fitting function, and use it as the correction value of the bad point.

3. A method for ultra-short-term prediction of equipment parameters in an integrated energy system according to claim 1, characterized in that: The parameters a, b i , h 1 (k - 1), h 2 The specific calculation expressions are as follows: U = (B T B) -1 B T Y where U represents a matrix of parameters, B T represents the transpose matrix of matrix B, (B T B) -1 represents the inverse matrix of matrix B T B; 4. An ultra-short-term prediction device for equipment parameters in an integrated energy system, characterized in that: It includes the following modules: A data acquisition module for collecting historical time-series data of each equipment parameter including the prediction parameter in the integrated energy system equipment and constructing a historical database; A mechanism analysis module for selecting characterization parameters related to the prediction parameter from the equipment parameters in the historical database according to the mechanism analysis of the equipment; The correlation analysis module is used to perform correlation analysis on the characterization parameters and extract feature parameters from the characterization parameters; The operating condition analysis module is used to calculate the characteristic identification quantity under different operating condition categories according to the feature parameters; The online prediction model construction module is used to construct an online prediction model for the integrated energy system equipment; The real-time prediction module is used to determine the operating condition category of the equipment to be predicted, select the extended sequence of the original basic sequence of the feature parameters that is most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, solve the online prediction model for the integrated energy system equipment, and obtain the predicted value of the equipment parameters of the equipment to be predicted; The performing correlation analysis on the characterization parameters and extracting feature parameters from the characterization parameters includes the following steps: 1) Obtain the prediction parameter x from the historical database 0 the reference sequence and the comparison sequences of p characterization parameters x 0 = {x 0 (1), x 0 (2), … x 0 (i) …, x 0 (n)} where x 0 is the reference sequence, and x 0 (i) is the i-th relevant parameter of the prediction parameter, is the comparison sequence of the i-th characterization parameter, is the j-th relevant parameter of the i-th comparison sequence, n is the number of elements included in the sequence, and p is the number of characterization parameters; 2) Perform dimensionless initialization processing on the reference sequence and the comparison sequence to obtain the dimensionless initialized reference sequence and comparison sequence; x' 0 = {1, x 0 (2) / x 0 (1), …, x 0 (n) / x 0 (1)} 3) Calculate the reference sequence x' after dimensionless initialization 0 and the comparison sequence on the k-th correlation parameter: where η is the resolution coefficient, represents the minimum difference between two levels, represents the maximum difference between two levels; 4) Calculate the grey relational grade γ of each characterization parameter i ; 5) Set the grey correlation degree threshold γ 0 , and extract the characterization parameters {γ i |γ i >γ 0 , i = 1, 2, …, p} to form a new sequence of characteristic parameters x 1 , x 2 , …, x q , where q is the number of characteristic parameters, q ≤ p; The calculating the characteristic identification quantity under different operating condition categories according to the feature parameters includes the following steps: Divide the operating condition categories according to the operating conditions of the equipment, and divide the historical data according to the operating condition categories; According to the divided historical data, calculate the characteristic identification quantity corresponding to the operating condition category; it includes the following steps: 1) Establish the characteristic parameter x 1 , x 2 , …, x q For the regression model of the prediction parameter x 0 ​ In the formula, is the regression coefficient, is the estimated value of the object calculated based on the regression model; 2) Obtain each regression coefficient from the following formula In the formula, are respectively the average values of x 0 , x 1 , x 2 , …, x q ; T and R are respectively the deviation matrix and the regression matrix of the regression model. where, T ij represents the element in the i-th row and j-th column of matrix T, and R i represents the i-th element in matrix R; 3) Characterize the prediction parameter x with λ 0 and the characteristic parameter x 1 , x 2 , …, x q The overall correlation degree under this working condition, and its calculation formula is as follows: Take λ as the characteristic identification quantity corresponding to each group of operating condition data; The constructing the online prediction model for the integrated energy system equipment includes the following steps: 1) When the prediction period is reached, starting from the current moment, simultaneously for the parameters x 0 , x 1 , x 2 , …, x q Trace back m historical data as the basic original sequence for online modeling: For perform an accumulative generation to obtain a sequence In the formula, 2) Generated by Generate the adjacent mean sequence In the formula, In the formula, α is the background value; 3) Establish the grey prediction model as where a is the development coefficient, and b i is the driving coefficient, h 1 (k - 1) is the linear correction amount; h 2 is the grey action amount; 4) The grey prediction model can be transformed into Wherein, 5) Perform a first-order accumulative reduction operation on the above formula to obtain the online prediction model of the prediction parameter: The determining the operating condition category of the equipment to be predicted includes the following steps: Using the original basic sequence, calculate the prediction parameter x 0 and the characteristic parameter x 1 , x 2 , …, x q the real-time identification quantity λ at the current moment; Determine the operating condition category of the equipment to be predicted according to the real-time identification quantity λ of the current working condition of the equipment and the determination condition; the determination condition is to satisfy the value of l below, where L is the total number of working condition classifications; The selecting the extended sequence of the original basic sequence of the feature parameters that is most similar to the current moment from the corresponding historical database as the input of the online prediction model for the integrated energy system equipment, solving the online prediction model for the integrated energy system equipment, and obtaining the predicted value of the equipment parameters of the equipment to be predicted includes the following steps: 1) In the corresponding historical operating condition data sequence, starting from the first historical data of the prediction parameter, select m data points to form a historical sequence By comparing with the measured sequence of the prediction parameter Perform similarity judgment to find the starting serial number point s of the similar sequence that satisfies the following formula in the historical data sequence; In the formula, is the data of the (s + k - 1)-th point in the historical sequence of prediction parameters starting from the s-th point and ending at the (s + m - 1)-th point; 2) Obtain the corresponding historical sequence of characteristic parameters Wherein, is the data of the (s + m + k - 1)-th point in the historical sequence of characteristic parameters starting from the (s + m)-th point to the end of the (s + m + f - 1)-th point, k is the number of points to be predicted, and f is the total number of predicted points; Integrate into the original basic sequence of characteristic parameters to obtain an extended sequence of the original basic sequence of characteristic parameters For perform an accumulation generation to obtain the input sequence of the prediction model Wherein, 3) Replace in the following formula and calculate 4) Substitute into the following formula to obtain the predicted values of the device prediction parameters at the next f points as follows:

Images